Font Size: a A A

Research And Design Of Layered Water-injection System Based On Improved Neural Network

Posted on:2015-08-29Degree:MasterType:Thesis
Country:ChinaCandidate:K ShiFull Text:PDF
GTID:2298330431995307Subject:Control theory and control engineering
Abstract/Summary:PDF Full Text Request
The main work of this paper is the research of the system design of layered waterinjection, the systerm can calculate the layered water injection accurately,each layereffectively used,make the production efficiency improved.Stratified water injection rate isforecast to be the core,including the prediction of the layered water injection,the pressurecalculation of the process of water injection,the calculation of the size of the stratified watermouth and the efficiency of the water injection system. An intelligent prediction software oflayered water injection is exploited according to the research results of this paper.The main contents include:1.the improved particle swarm optimization is combined with BP neural network: In thispaper,the learning factor is adjusted to optimize the global search ability and local searchability of the particle swarm algorithm In order to improve the precision and convergencespeed of the algorithm;and chaos is used to optimize the initial value of the particle swarm,which avoids the blindness of optimization and increases the ergodicity and stability ofparticles; inertia weight adaptively adjusted according to the degree of premature convergenceof the groups to ensure the algorithm’s converging to the global optimal.Two classicalfunctions are used to test the improved particle swarm optimization, and it’s showed by theexperimental results that the improved algorithm is feasible. Find the optimal solution byACPSO instead of BP neural network by error back propagation algorithm to update theweights and thresholds.2.Prediction of layered water injection:To determine the factors influencing the layeredwater injection,the predictive model of layered water injection based on the improved particleswarm neural network is constructed,The influencing factors as the input layer,layered waterinjection as the output layer,training the model,to predict the layered water injection with thetrained model. a sensitivity analysis is done to the reservoir parameters which is used to verifythe accuracy of the forecast data.3.Design for the water injection system: according to the calculation formula of waterinjection pressure, the prediction of water injection into the income of each layer, which isused to predict the water injection pressure; through the cycle of trial way to tap thedimension is calculated, thus avoiding tap tap size inaccuracy caused by repeated fishing;through calculation and increase the injection efficiency, see the effect of injection and perfectsystem design.
Keywords/Search Tags:BP neural network optimization, Adaptive chaotic particle swarmoptimization algorithm, intelligent prediction of layered injection
PDF Full Text Request
Related items